Programme archive: |
| »DRBD – building shared nothing HA clusters« * |
| , LinBit |
 |
| DRBD and Heartbeat allow you to build cost efficient high availability (HA) clusters using the Linux operating system and PC server hardware. rnIn an HA cluster you have two server nodes that are capable of running the service that you want to have ensuring high availability. The cluster management software (Heartbeat) monitors the nodes in your cluster and takes care of cluster resource allocation and of starting your services.rnDRBD is the missing piece to have dynamic data replicated over the two nodes in your HA cluster. This allows you to also have services like databases (PostgreSQL, Oracle, DB2, ...), fileservers (Samba, NFS) in your HA clusters.rnThis work will point out all the nifty details of such an HA cluster setup. You will get the insight on how resynchronisation of disks will occur after (partial) cluster failures. How to estimate performance and resynchronisation times, how to adjust Heartbeat and to tune DRBD to the specific problem. rnThese HA clusters can be used in a broad range of systems from data-centre like setups (operator is around the corner) to embedded systems. |
| |
|
|
